CN101515816B - Self-adapting subgroup precoding method of maximal SJNR criterion - Google Patents

Abstract

The present invention relates to the field of wireless mobile communication. Aiming at the problem that high performance and low complexity cannot coexist in precoding technology, and the suppression effect on noise will be worse when the noise variance is small, the present invention is based on the principle of maximizing the SJNR criterion Grouping precoding algorithm, an adaptive grouping precoding scheme is designed. This method first adopts the grouping technology. In the multi-user multiple-input multiple-output system (MIMO), the CSI information is obtained from the upper layer, and the adaptive interference and noise suppression is performed according to the CSI. The base station divides multiple users into several groups, and the transmission channel The CSI situation judges the channel state. When the channel state is poor, the group precoding that maximizes the SJNR criterion is used to suppress interference and noise. When the channel state is good, the group joint transmission algorithm is directly used. This adaptive group precoding algorithm can achieve the best bit error rate performance under different channel conditions. The present invention can be widely applied in MIMO systems.

Description

Adaptive group precoding method for maximizing SJNR criterion

technical field

The invention relates to the field of wireless communication, in particular to wireless transmission preprocessing technology.

Background technique

At present, the receiving/detection technology is becoming more and more complex, which puts great pressure on the receiver (mobile terminal) of the mobile communication downlink. Some experts and scholars have begun to pay attention to and explore the "moving" of the complex receiving/detection algorithm to the sending end. (base station) to reduce the complexity and power consumption of the receiving end, so that the detection technology can be placed on the sending end for preprocessing, thereby forming a preprocessing technology, also known as precoding (Precoding) technology (also known as joint transmission (JT )), which greatly simplifies the complexity of the downlink receiver.

.JointTransmission(JT)，an alternative rationale for the downlink of TimeDivision CDMA using multi-element transmit antennas.Proc.IEEE SixthInternational Symposium on Spread Spectrum Techniques and Applications(ISSSTA 2000)，Parsippany，2000，pp：1-5.]首先给出了在多入单出(MISO)系统中的联合传输技术(JT)，该技术以联合检测(JD)技术为原理，但将其过程反过来，把接收端的检测工作放在基站处进行了预处理，大大简化了接收端的复杂度，而且还增强了系统性能。文献2[海平，谢显中.联合传输技术(JT)及其算法简化.通信学报，2003，24(11A)，pp：93-100]中研究了几种对JT技术的简化算法，主要针对系统矩阵的算法进行了简化，降低了计算复杂度。文献3[冯媛，谢显中.降低多用户MIMO下行检测复杂性的联合发送技术，电子信息学报，2007.1：174-176.]将JT技术扩展到多入多出(MIMO)系统中，形成JT-MIMO技术，并进一步讨论比较了JT-MIMO算法与JT、JD算法的优劣。以上3个文献主要是从预均衡的角度出发进行预编码设计，而且采用的是迫零准则，该准则的特点是能消除符号间干扰和多址干扰，但也增强了噪声。文献4[Yongle Wu，Jinfan Zhang，Mingguang Xu，Shidong Zhou，Xibin Xu，Multiuser MIMO Downlink Precoder Design Based on the MaximalSJNR Criterion，IEEE GLOBECOM’05，St.Louis，USA，28 Nov.-2 Dec.2005，vol.5，pp：2694-2698.]从平衡共信道干扰抑制和噪声抑制的角度出发，设计了SJNR(Signal to Jamming and Noise Ratio)准则，利用该准则，通过在每用户发射功率约束下最大化SJNR，将其最优解作为预编码器。与迫零准则相比，基于SJNR的预编码提高了系统性能，对天线数也没有限制，但是，在噪声方差小的时候对噪声的抑制效果会变差，当用户数较多时计算复杂度较高。Literature 1 [PWBaier, M.Meurer, T.Weber, H.

.JointTransmission(JT), an alternative rationale for the downlink of TimeDivision CDMA using multi-element transmit antennas.Proc.IEEE Sixth International Symposium on Spread Spectrum Techniques and Applications(ISSSTA 2000), Parsippany, 2000, pp: 1-5.] First The joint transmission technology (JT) in the multiple input single output (MISO) system is given. This technology is based on the joint detection (JD) technology, but the process is reversed, and the detection work at the receiving end is carried out at the base station. The preprocessing greatly simplifies the complexity of the receiving end, and also enhances the system performance. Document 2 [Hai Ping, Xie Xianzhong. Joint Transmission Technology (JT) and Its Algorithm Simplification. Journal of Communications, 2003, 24(11A), pp: 93-100] studied several simplified algorithms for JT technology, mainly for the system matrix The algorithm is simplified to reduce the computational complexity. Document 3 [Feng Yuan, Xie Xianzhong. Joint transmission technology to reduce the complexity of multi-user MIMO downlink detection, Journal of Electronic Information, 2007.1: 174-176.] extended JT technology to multiple input multiple output (MIMO) system, forming JT- MIMO technology, and further discussed and compared the advantages and disadvantages of JT-MIMO algorithm and JT, JD algorithm. The above three documents mainly carry out precoding design from the perspective of pre-equalization, and adopt the zero-forcing criterion. The characteristic of this criterion is that it can eliminate inter-symbol interference and multiple access interference, but it also enhances noise. Literature 4 [Yongle Wu, Jinfan Zhang, Mingguang Xu, Shidong Zhou, Xibin Xu, Multiuser MIMO Downlink Precoder Design Based on the MaximalSJNR Criterion, IEEE GLOBECOM'05, St.Louis, USA, 28 Nov.-2 Dec.2005, vol .5, pp: 2694-2698.] From the perspective of balancing co-channel interference suppression and noise suppression, the SJNR (Signal to Jamming and Noise Ratio) criterion is designed. Using this criterion, by maximizing the SJNR, using its optimal solution as a precoder. Compared with the zero-forcing criterion, the precoding based on SJNR improves the system performance, and there is no limit to the number of antennas. However, when the noise variance is small, the noise suppression effect will be worse, and when the number of users is large, the computational complexity is relatively high. high.

Contents of the invention

The present invention aims at the problem that high performance and low complexity cannot coexist in the precoding technology, and the noise suppression effect will be worse when the noise variance is small. State information (CSI), a cluster precoding algorithm based on maximizing the SJNR criterion is proposed, and an adaptive cluster precoding scheme is designed on this basis, thereby forming an adaptive cluster precoding algorithm maximizing the SJNR criterion.

The technical solution of the present invention to solve the above technical problems is to design an adaptive grouping precoding method that maximizes the SJNR criterion. The base station groups the users in its coverage area, and can use dynamic grouping or static grouping to group users. The channel state information CSI of the transmission channel is obtained from the upper layer. When the channel state information CSI ≥ the critical value Δ, the grouping joint transmission algorithm is directly used to preprocess the transmission signal; when the channel state information CSI is smaller than the critical value Δ, the maximum value is used. The SJNR criterion preprocesses the transmitted signal. The grouping joint transmission algorithm specifically includes, according to the spreading matrix C _g and the demodulation matrix D _g to establish a matrix B _g =D _g H _g , and the matrix B _g is based on the zero-forcing criterion M _g =B _g ^H (B _g B _g ^H ) ^-1 obtains the modulation matrix M _g , and modulates the data d _G ^g of each group through its own modulation matrix M ¹ _G ... M ^g _G . The maximum SJNR criterion is the combination of the modulator, demodulator and channel that maximizes the SJNR to form the SJNR preprocessing module. According to the maximum eigenvalue λ _max of (H _G ^k F _G ) ^H H _G ^k F _G , call formula $f_G=R_G\×h_G^g\×f_G=\sqrt{{\mathrm{\λ}}_{\max }}$ Determine the SJNR gain factor f _G , after the data d _G ^g of each group is modulated by their own modulation matrix M _g ^G , the SJNR gain factor f _G is used to maximize the SJNR preprocessing.

The present invention judges the channel state for the CSI condition of the transmission channel, adopts the group precoding that maximizes the SJNR criterion to suppress interference and noise when the channel state is poor, and directly adopts the group joint transmission algorithm when the channel state is good.

This algorithm can not only reduce the number of channels to be processed by grouping processing, but also obtain a lower system complexity; it can also adaptively select a transmission processing algorithm according to the quality of the channel state, suppress interference and noise, and reduce the bit error rate ; Further, the grouped signals pass through the same channel and are transmitted through two (or more) transmitting antennas, which has the effect of diversity and multiplexing, and also improves system performance to a certain extent.

Description of drawings

Fig. 1 main implementation steps of the present invention

Figure 2 Grouping joint transmission system model

Figure 3 Maximize the SJNR cluster precoding system model

Figure 4 SJNR gain factor f _g structure diagram

Figure 5 Performance comparison of joint transmission (JT), group joint transmission (GJT), SJNR grouping algorithm (GSJNR), and adaptive grouping scheme (SELECT)

Detailed ways

Embodiments of the present invention will be specifically described below in conjunction with the accompanying drawings. For the convenience of description, the present invention considers the TDD-SCDMA system based on TDD (Time Division Duplex), and implements in other systems in a similar manner.

Figure 1 is the main implementation steps of the present invention, which specifically includes: first, the base station groups users in its coverage area into groups, the grouping methods include dynamic grouping and static grouping, and the joint transmission algorithm is used to process the data of each group separately, from The upper layer obtains the channel state information CSI of the transmission channel. For the TD-SCDMA system, the base station obtains the channel state information CSI of the transmission channel required for downlink precoding through uplink channel estimation. For the FDD wireless system, the CSI is obtained through feedback. Then, adaptive interference and noise suppression is performed according to CSI.

The maximum SJNR criterion can achieve better bit error rate performance than group joint transmission when the channel state is poor. Below the critical value, the signal should be transmitted to maximize SJNR processing to reduce the bit error rate. If it is above the critical value, no processing is required, and group joint transmission can be performed directly. Therefore, we first judge the channel state according to the specific channel conditions, and pre-set a critical value Δ. The specific setting of the critical value is to select a system-acceptable signal-to-noise ratio as the critical point by means of anti-true experiments. To set the critical value Δ, in the following simulation conditions in this paper, the SNR=-1 is selected as the critical point through experiments, and the critical value Δ=-1 is set. When the channel state information is below this value (CSI<Δ), the maximum SJNR criterion is used to preprocess the transmission signal; if the channel state information is higher than the critical value (CSI≥Δ), the grouping joint transmission algorithm is directly used , to preprocess the transmitted signal. That is to say, according to the channel state information CSI, the channel state is judged on the transmission channel condition. When the channel state is poor, the group precoding that maximizes the SJNR criterion is used to suppress interference and noise. When the channel state is good, group joint transmission is used. The system model directly preprocesses the signal by grouping joint transmission algorithm.

A grouping joint transmission system model is established to group the users within the coverage of the base station to reduce the number of channels to be processed and obtain a lower system complexity. Two methods can be used to realize user grouping. The first is dynamic grouping. The system sets the number of groups according to the maximum number of users supported, and the number of users in each group changes dynamically, and detects the number of users currently receiving services in real time. The number of users served is divided into multiple groups on average; the second is static grouping, and the number of users in each group remains unchanged. First, the number of users in each group is fixed according to the processing capacity of the system, and the serving users are evenly divided into multiple groups. After grouping, the data of each group is processed separately using a joint transmission algorithm, which is called group joint transmission.

The processing method of the joint transmission algorithm is specifically described as follows. Figure 2 shows the grouping joint transmission system model. Assume that the system has K users. For example, all users are divided into G groups on average according to the static grouping principle, and each group has L users. The data d _G ^g (g=1...G) of each group is sent to the modulation matrix as an independent data for processing. Assuming that the spreading matrix corresponding to each group is C _g , and the demodulation matrix is D _g , the present invention takes ${\mathrm{D.}}_g=C_g^h$ (The system here is TD-SCDMA, if it is a non-spread spectrum system, D _g can be selected as a matched filter), and the channel matrix H _g (the dimension of H _g is determined by the number of users in each group). A matrix B _g is established according to the spread spectrum matrix C _g and the demodulation matrix D _g , and the modulation matrix M _g =B _g ^H (B _g B _g ^H ) ^-1 is obtained from the matrix B _g according to the zero-forcing criterion. The original data d _G ^g of each group of users is modulated by their respective modulation matrices M ¹ _G ... M ^g _G , and then data transmission is performed through their respective channels H _g , and the receiving end receives the data of each channel The modulated data are respectively demodulated by respective demodulators (D _g ¹ ... D _g ^G ) to obtain mutually independent group estimation signals.

When the channel state information CSI<Δ, the maximum SJNR criterion is used to preprocess the transmission data at the base station to maximize the SJNR. Figure 3 shows the maximum SJNR group precoding system model. After the grouped data d _G ^g (g=1...G) of each group is modulated by the modulation matrix M _g ^G (g=1...G), the modulated signal t _G ^g (g=1...G) is obtained and sent to the SJNR respectively The pre-processing module f _g (g=1...G) performs maximum SJNR pre-processing through the SJNR gain factor, and then transmits through the channel H _G ^g . At the receiving end, correspondingly, the SJNR demodulator D _g is firstly used to demodulate the received signal and restore it to the original signal.

The SJNR criterion measures the impact of a user on the entire system, and the expression for determining the maximum SJNR is as follows:

$\underset{{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}{\mathrm{<span\; class="notranslate">\; max</span>}}{\mathrm{<span\; class="notranslate">\; SJNR</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}{\mathrm{<span\; class="notranslate">\; h</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}{\mathrm{<span\; class="notranslate">\; h</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}{{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&NotEqual;</span>\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; K</span>}}{\mathrm{<span\; class="notranslate">\; h</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}{\mathrm{<span\; class="notranslate">\; h</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="0"\; label="N"\; filenames="H2009101035187E00031.png"\; state="\{\{state\}\}">N</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}}$

$<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; J</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}{{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&NotEqual;</span>\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; K</span>}}{\mathrm{<span\; class="notranslate">\; h</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}{\mathrm{<span\; class="notranslate">\; h</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="0"\; label="N"\; filenames="H2009101035187E00031.png"\; state="\{\{state\}\}">N</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}}$

st $t_k^h{t}_k=P_k,$ $J_k={\mathrm{\&Sigma;}}_{i=1,i\&NotEqual;k}^K{t}_k^h{h}_i^h{h}_k{t}_k,$ k=1...G

Among them, P _k is the transmission power of the kth user, J _k represents the interference power of user k to other users, t _k and H _k represent the modulated signal of user k and the passed channel, N ₀ represents the variance of noise.

It can be seen that each user will interfere with other users, and the interference J _k is independent of each other, changing the transmit power of any user will not affect the J _k of other users, so it is easy to realize mathematically. Just because J _k are independent of each other, each user has its own SJNR, so the optimization of SJNR should also be carried out for each user. The optimal solution to maximize SJNR is the formula ${\mathrm{\&Sigma;}}_{i=1,i\&NotEqual;k}^K{h}_i^h{h}_i+\frac{N_0}{P_k}I$ (Here I is the maximum eigenvector of the unit matrix), and this maximum eigenvector is set as the modulator F _k of maximizing the SJNR, expressed as follows:

${\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\sqrt{{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; \&zeta;</span>}}_{\mathrm{<span\; class="notranslate">\; m</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; Y</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>$

${\mathrm{<span\; class="notranslate">\; Y</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&NotEqual;</span>\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; K</span>}}{\mathrm{<span\; class="notranslate">\; h</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}{\mathrm{<span\; class="notranslate">\; h</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; +</span>\frac{{\mathrm{<span\; class="notranslate">\; N</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}}{{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; )</span>}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; h</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}{\mathrm{<span\; class="notranslate">\; h</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}$

Here, ζ _m (·) represents the largest eigenvector corresponding to the largest eigenvalue taking the value in brackets. In this way, the corresponding demodulator R _k adopts the form of matched filtering, namely $R_k=\frac{{\left(h_k{f}_k\right)}^h}{\left|\right|h_k{f}_k\left|\right|},$ Where ‖·‖ means 2 paradigms.

Maximizing the SJNR criterion means maximizing the SJNR group precoding system model, combining the SJNR modulator F _G , demodulator R _G and channel H _G into one to form the SJNR factor, eliminating the need for the transmitted signal to go through SJNR modulation and decoding. The adjustment process simplifies the processing process. In order to simplify the process, F _k , H _k and R _k can be combined into one part, which is set as the SJNR gain factor f _k , where k=1...G. Figure 4 is a schematic diagram of the SJNR gain factor. According to the formula:

$f_G=R_G\&times;h_G^g\&times;f_G=\frac{{\left(h_G^k{f}_G\right)}^h}{\left|\right|h_G^k{f}_G\left|\right|}\&times;h_G^k\&times;f_G=\sqrt{{\mathrm{\&lambda;}}_{\max }}$ (k=1...G), determine the SJNR gain factor. Among them, λ _max represents the maximum eigenvalue of (H _G ^k F _G ) ^H H _G ^k F _G . After the SJNR gain factor is determined, the transmitted signal is directly multiplied by the corresponding SJNR gain factor f _G to realize the maximum SJNR preprocessing of the signal.

Fig. 5 is the performance comparison of Joint Transmission (JT), Grouping Joint Transmission (GJT), SJNR Grouping Algorithm (GSJNR), and Adaptive Grouping Scheme (SELECT). In the TD-SCDMA system, when there are 6 users communicating at the same time in a certain time slot (users in different time slots can be considered as having no interference with each other), the static grouping method is adopted, and the SNR=-1 is selected as the critical point, and the simulation environment is Parameters of multipath fading channel case3 of TD-SCDMA in 3GPP. It can be seen that the bit error rate of the three algorithms is lower than that of the joint transmission, and the adaptive clustering algorithm is close to the bit error rate of the SJNR clustering algorithm when the signal-to-noise ratio is low, and is close to the bit error rate of the grouping joint transmission when the signal-to-noise ratio is high , can indeed take the advantages of these two algorithms, and guarantee a better bit error rate performance on the whole.

Claims (3)

1. The adaptive grouping precoding method of maximizing the SJNR criterion is characterized in that the base station divides the users in its coverage area into groups, and obtains the channel state information CSI of the transmission channel from the upper layer, when the channel state information CSI is greater than or equal to the critical value Δ , according to spreading matrix C _g and demodulation matrix D _g to establish matrix B _g ＝ D _g H _g , and obtain modulation matrix M from matrix B _g according to zero-forcing criterion M _g ＝ B _g ^H (B _g B _g ^H ) ^-1 _g , take the data d _G ^g of each group of users as an independent data and modulate them through their respective modulation matrices M ¹ _G ...M ^g _G , where g=1...G, G is the number of user groups ; When the channel state information CSI is less than the critical value Δ, the maximum SJNR criterion is used to preprocess the transmission signal, that is, the SJNR modulator F _G , demodulator R _G and channel H _G are combined into one to form the SJNR gain factor ,according to

The largest eigenvalue λ _max of the call formula Determine the SJNR gain factor f _G , each group of data is modulated by its own modulation matrix, and the modulated signals are sent to the SJNR preprocessing module to maximize the SJNR preprocessing through the SJNR gain factor f _G , where k=1...G. 2. adaptive grouping precoding method according to claim 1, it is characterized in that, to TD-SCDMA system, base station obtains the channel state information of the transmission channel that downlink precoding needs by uplink channel estimation, for FDD wireless system, its CSI is obtained through feedback. 3. The adaptive grouping precoding method according to claim 1, wherein grouping users includes dynamic grouping and static grouping, and the dynamic grouping is that the system sets the number of groups according to the maximum number of users supported, The number of users in each group changes dynamically, and the number of users currently receiving services is detected in real time, and they are divided into each group on average; the static grouping is to fix the number of users in each group according to the processing capacity of the system, and divide the users being served evenly into each group.

Images